Monster VM Best Pracitces.

[vmtech123]

Hello, as far as I am aware this is not created yet.

Is there a chance that this could be done, or could people who back up large VM's, (5-30TB) share what they have done to keep times down?

I have created the SAN volumes, Datastores, and am planning on splitting up my VMDK's into 5TB disks to keep the sizing down and just add 5TB at a time going forward. 2TB is going to create too much overhead to do this all time time.

When I send my jobs to tape, obviously this will be a single job, but I am getting a new library with 8 drives. I will also make sure create many tape jobs for my "monster" file servers (I have a lot) so they can backup at the same time. Yes, I won't be able to use parallel drives for restores, but I rarely restore from tape, and if I do, it's because someone wants a file that has been purged out of disk backup. In this case it's been months usually and they can wait. Tape is mostly for airgap, offsite, disasters.

I would love to see more information with monster VM's, especially some real world testing / comparisons. Splitting the VM's between datastores, different sized disks etc. Or does it just make more sense to back up a VM with 2 15TB disks and call it a day?

Thanks

[foggy]

I wouldn't recommend splitting the VM into several per-disk jobs if that is what you're thinking about. This could cause issues during FLR. Overall, people do backup VMs that large, no issues with that. The more you're able to parallelize, the better backup times you will have, so having several disks is of course recommended.

[HannesK]

Hello,
sure, you can back up large VMs. 5-30TB sounds relatively easy compared to the 640 TB VM I heard about a few weeks ago.

Multiple VMDKs is good for more parallel tasks. 5TB sounds good. The 640TB VM uses up to 20TB per virtual disk.

You might also be interested in backup from storage snapshots: https://helpcenter.veeam.com/docs/backu ... ml?ver=110

Multiple datastores might not be required with that size of VM, as VMware allows 62TB datastores. It's hard to predict and also depends on your environment.

Best regards,
Hannes

[Andreas Neufert]

Based on my past field experience I had a rule of thumb for those VMs.
If you can stay below 10 virtual disks. Each disk add some time for snapshot commits and the more disks the more overhead.
On the other side, the more VM disks the more Veeam parallelization you will get and so more backup speed.
10 VM disks ar a sweet spot.

For NTFS disks stay below 64TB so that chkdsk still works.